tag 613348 unreproducible moreinfo severity 613348 important thanks Hello Toni,
sorry for the late response, cleaning up old bug mail. Toni Mueller [2011-02-14 10:22 +0100]: > after running pg_upgradecluster to upgrade from 8.3 (Lenny) to 8.4 > (Squeeze), my pg_hba.conf in the 8.3 release was nuked The evidence is probably long gone, but do you still remember what "nuked" meant there? Was it an empty file? No file at all? Was there a pg_hba.conf.orig? > I see no reason why the program should mangle the old pg_hba.conf in the > first place. The man page also gives no warning whatsoever. It's temporarily changed to ensure that only the local superuser has access to it during upgrade. I got other bug reports in the past that other apps tried to access the DB while the upgrade was running, which either disrupted the upgrade or at least led to inconsistencies. I just looked at the code for creating the backup and temporary pg_hba.conf, and it indeed could be made a tad more robust. But I don't see how it can create an empty pg_hba.conf without renaming the original file as pg_hba.conf.orig first. pg_upgradecluster never removes that file, it just uses rename(). Do you happen to have a way to reproduce this? Thanks, Martin -- Martin Pitt | http://www.piware.de Ubuntu Developer (www.ubuntu.com) | Debian Developer (www.debian.org) -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
